AQUATIC BIOLOGYAquat Biol

Vol. 19: 111–127, 2013doi: 10.3354/ab00527

Published online September 24

INTRODUCTION

The growing human population has severe envi-ronmental impacts, often summarised under the term‘global change’. In the marine realm, 2 of the mainforces causing significant changes are global warm-ing and ocean acidification (OA), both largely drivenby the burning of fossil fuels that has led to the cur-rent dramatic rise in carbon dioxide partial pressure(pCO2) and increasing sea surface temperatures.Coral reefs are perceived as one of the main victimsof this development, where warming and OA poten-tially have significant consequences for the calciumcarbonate (CaCO3) balance (Hoegh-Guldberg et al.2007, Erez et al. 2011, Veron 2011). While the com-

plex effects of temperature may both promote orinterfere with coral calcification (Reynaud et al. 2003,Rodolfo-Metalpa et al. 2011), OA was demonstratedto have chiefly negative effects on health, reproduc-tion, growth, and survival of corals and other marinebiota (De’ath et al. 2009, Kroeker et al. 2010, 2013,Andersson et al. 2011, Pandolfi et al. 2011). Never-theless, some organisms and processes appear tobenefit in a high-CO2 world, as evidenced for in -stance by greater cover of non-calcareous macro-phytes and the massive coral Porites—but also a sig-nificant increase of internal macrobioerosion withinthe latter species at a Papua New Guinean coral reefwith a naturally reduced pH (Fabricius et al. 2011).Bioerosion is a key process in the carbonate budget

© The authors 2013. Open Access under Creative Commons byAttribution Licence. Use, distribution and reproduction are un -restricted. Authors and original publication must be credited.

Publisher: Inter-Research · www.int-res.com

*Email: max.wisshak@senckenberg.de

Effects of ocean acidification and global warmingon reef bioerosion—lessons from a clionaid sponge

Max Wisshak1,*, Christine H. L. Schönberg2, Armin Form3, André Freiwald1

1Senckenberg am Meer, Marine Research Department, 26382 Wilhelmshaven, Germany2Australian Institute of Marine Science (AIMS), University of Western Australia, Crawley, Western Australia 9006, Australia

3GEOMAR Helmholtz Centre for Ocean Research, Marine Biogeochemistry, 24105 Kiel, Germany

ABSTRACT: Coral reefs are under threat, exerted by a number of interacting effects inherent tothe present climate change, including ocean acidification and global warming. Bioerosion drivesreef degradation by recycling carbonate skeletal material and is an important but understudiedfactor in this context. Twelve different combinations of pCO2 and temperature were applied to elu-cidate the consequences of ocean acidification and global warming on the physiological responseand bioerosion rates of the zooxanthellate sponge Cliona orientalis—one of the most abundant andeffective bioeroders on the Great Barrier Reef, Australia. Our results confirm a significant amplifi-cation of the sponges’ bioerosion capacity with increasing pCO2, which is expressed by more car-bonate being chemically dissolved by etching. The health of the sponges and their photosymbiontswas not affected by changes in pCO2, in contrast to temperature, which had significant negativeimpacts at higher levels. However, we could not conclusively explain the relationship betweentemperature and bioerosion rates, which were slightly reduced at both colder as well as warmertemperatures than ambient. The present findings on the effects of ocean acidification on chemicalbioerosion, however, will have significant implications for predicting future reef carbonatebudgets, as sponges often contribute the lion’s share of internal bioerosion on coral reefs.

KEY WORDS: Global change · Carbon dioxide · Temperature · Bioerosion · Cliona orientalis · Great Barrier Reef

OPENPEN ACCESSCCESS

Aquat Biol 19: 111–127, 2013

of coral reefs, yet it is understudied. It acts at variousscales and is performed by a multitude of organismsencompassing chemical and/or mechanical meansduring superficial grazing, attachment, or internalmicro- and macrobioerosion (e.g. Wisshak & Tapa -nila 2008, Tribollet & Golubic 2011, Tribollet et al.2011). While healthy reefs maintain a positive budgetor at least equilibrium between calcification and bio-erosion, certain environmental conditions can tip thescale towards erosion (e.g. Glynn 1997, Perry et al.2008). Acknowledging the importance of bioerosionfor structuring coral reefs, recent experimental stud-ies have looked at the effects of OA and confirmed anacceleration of bioerosion with increasing pCO2 for2 principal agents of internal bioerosion: phototrophicmicrobial euendoliths (Tribollet et al. 2006, 2009,Reyes-Nivia et al. 2013) and bioeroding sponges(Wisshak et al. 2012). These studies considered OAas the sole factor or used fixed scenarios preventingseparate evaluation of environmental factors, and westill cannot answer how global warming may sup-port, balance, or counteract increasing rates of reefbioerosion. Different scenarios may be possible fortemperature effects on chemical bioerosion (i.e. bio-genic dissolution of CaCO3), because on the onehand, the speed of a biochemical reaction is directlyproportional to temperature (Atkins & De Paula2009), but on the other hand, the capacity of physico-chemical dissolution of CaCO3 in seawater is nega-tively proportional to temperature (Mehrbach et al.1973), possibly counteracting the effects of OA onbioerosion (Wisshak et al. 2012). In the present study,we report observations on sponge bioerosion affectedby OA, global warming, and their interactions, re -sulting from an experiment involving 12 differentcombinations of temperature and pCO2.

The bioeroding organism chosen for the experi-ment was Cliona orientalis Thiele, 1900, a dominantand rapidly bioeroding, zooxanthellate sponge andmember of the ‘Cliona viridis species complex’ whichis ubiquitous on tropical coral reefs worldwide (Schön -berg 2000, 2002a,b, van Soest et al. 2013). C. orien-talis may reach several square metres in colony sur-face area at over 1 cm of penetration depth, androutinely invades and kills live corals (Schönberg2000, Schönberg & Wilkinson 2001). C. orientalis wasreported to survive and increase in abundance after aheating event causing severe coral bleaching (Schön -berg & Ortiz 2009), and it is particularly robust andoverall very suitable for experimental studies (Schön-berg & Wilkinson 2001, Schönberg 2002b, Wisshak etal. 2012). C. orientalis erodes by biochemical dissolu-tion, leading to the formation of cup-shaped grooves

and the mechanical extraction of so-called spongechips (Schönberg 2008), a very effective mode ofpenetration resulting in annual bioerosion rates of>10 kg CaCO3 m−2 of calcareous substrate for thisspecies (Schönberg 2002b). Within the local commu-nity of bioeroding sponges at the Great Barrier Reef,C. orientalis is the most abundant species withrespect to the infested reef surface area (Schönberg& Ortiz 2009), and owing to its high rate of bioerosion(Schönberg 2002b), it is also the most destructiveone, emphasising the relevance of potential effects ofclimate change on the bioerosion rate of this species.

MATERIALS AND METHODS

Sampling and maintenance of Cliona orientalis

All sampling and experimental procedures wereconducted at Orpheus Island Research Station (OIRS),located on Orpheus Island, central Great BarrierReef, Australia (Fig. 1). The local community of bio-

112

Fig. 1. Location map of Queensland, with Orpheus Island(Palm Island Group) on the central Great Barrier Reef (top).The sample site in Little Pioneer Bay is just north of the

Orpheus Island Research Station (OIRS)

Wisshak et al.: Global change impacts on sponge bioerosion

eroding sponges is diverse, comparatively well stud-ied (e.g. Schönberg 2000, 2001), and has recently in -creased in abundance, possibly as a result of declin-ing reef health related to a heating event resultingin widespread coral death (Schönberg & Ortiz 2009).At the leeward fringing reef, C. orientalis was core-sampled in 1 to 3 m water depth from massive Poritessp. colonies with an air-drill and hole-saw (inner dia -meter: 30 mm), and later trimmed to 25 mm in lengthwith an air-cutter, so that cores included sponge-penetrated material in the upper half and clean coralskeleton below (see Fig. 1D in Wisshak et al. 2012).Individual variation in the amount of sponge tissue inC. orientalis cores was assumed to be insignificant, asevident from the very even erosion pattern in thisspecies (Schönberg & Shields 2008) and the low stan-dard deviations both for penetration depth as well assponge biomass measured in 128 sponge-bearingcores taken from 10 different C. orientalis colonies inLittle Pioneer Bay (see Table 1 in Wisshak et al.2012). Cores were kept for 4 d in an outdoor flow-through tank for recovery, allowing tissue to fully healbefore transferring them to a constant-temperatureroom, offering an acclimatisation period of 24 h. Ac -climatisation and experiments were conducted intotal darkness to avoid fluctuations in pH caused byalterations of photosynthesis and respiration due tolight/dark cycling and variations in temperaturecaused by lamps (Wisshak et al. 2012). Directly be -fore and after the experiment, the condition of thesponge cores was documented with digital photo -graphy including a colour chart. Photographs werevisually examined, assessing images on greyscaleand colour, and scoring them as follows: 10: darkchocolate brown, 9: milk chocolate brown (normalbefore dark adaptation), 8: toffee brown, 7: brown-ochre (normal after dark adaptation), 6: ochre/yel-low-ochre, 5: pale ochre, 4: very pale to translucent,3: with patches of black disease, likely to recover,2: most tissue black, but with small portions alive,unlikely to recover, 1: recently dead, black withwhite film, 0: dead, black, tissue mostly disinte-grated. Thus we recorded paling when symbiontswere shifted into deeper tissue layers, a processwhich is affected by light changes, but also by stressand other discolouration (Schönberg & Suwa 2007),with special attention given to tissue death. The photosymbiosis between the sponge and its dino -flagellate symbionts was as sumed to be vital to thesponges’ health and physio logical functions (e.g.Rosell 1993, Weisz et al. 2010, Hill & Hill 2012), andphotosynthetic parameters were monitored in thedark-adapted samples at the beginning and at the

end of the experiment, using pulse-amplitude modu-lated fluorometry (PAM; Maxi-iPAM, Walz). Photo-synthetic efficiency or quantum yield Fv/Fm (ratio ofvariable to maximum photosynthetic fluorescence)provided a proxy for the health of the symbionts,while baseline photosynthetic fluorescence F0 wasused as a proxy for chlorophyll concentration, allow-ing us to recognise whether symbionts were shifteddeeper into the sponge cores, which can be a reac-tion to stress (e.g. Schönberg & Suwa 2007).

Experimental setup

All experiments were performed under constantconditions under which the sponge-bearing coreswere subjected to different standardised measure-ments and sampling protocols in a semi-closed sys-tem (Fig. 2). To establish consistent water quality andproperties throughout the experiment, ambient sea-water was filtered to 5 µm (thereby retaining pico -plankton <3 µm as the main source of nutrition, seee.g. Lynch & Phlips 2000), and stored in a 600 l reser-voir tank connected to a pump and UV unit. Stableexperimental temperatures were achieved throughclimate control of the entire room.

A series of 4 independent 72 h experiments wererun at ~22°C (colder than ambient), ~25°C (ambient),~28°C (warmer than ambient), and ~31°C (stress tem -perature). Each set of temperatures was run against3 levels of pCO2 (present-day: ca. 400 µatm, moder-ately elevated: ca. 750 µatm, and strongly elevated:ca. 1700 µatm), resulting in a treatment matrix with12 different combinations of temperature and pCO2.The different pCO2 levels were achieved by pertur-bation with gases specifically mixed with Digamix5KA 36A/9 pumps on M755-1.2 consoles (H. Wöst -hoff) using food-grade CO2 and compressed air withambient pCO2, and only compressed air for the present-day pCO2.

Each treatment included 4 replicate bioreactors(graded 3 l beakers), each with 8 sponge-bearingcores sitting on a tethered plastic grid in an aliquot of2.5 l of pre-filtered seawater taken from the reservoirtank. Since donor effects can be caused by geneticdifferences of the sampled sponges (Maldonado etal. 2008), we attempted the best possible spread ofdonor diversity per bioreactor, with cores from atleast 6 specimens per set. Due to sampling restric-tions, we were unable to include true controls (cleancoral cores), but we had demonstrated earlier thatover such a short period of time, potential abioticchemical dissolution of the substrate or possible bio-

113

Aquat Biol 19: 111–127, 2013

erosion by microendoliths remains negligible, evenunder strongly elevated pCO2 (Wisshak et al. 2012).However, in each treatment, one additional bioreac-tor with seawater only (blank) served the purpose ofmonitoring the water chemistry. The temperature inthe bioreactors was kept very stable by placingthem into 150 l temperature buffer tubs with wateradjusted to the constant room temperature. Uniformperturbation per bioreactor proceeded via a flowcontroller that pumped the gas mixes into the bio -reactors at fixed, equal rate per bioreactor, and inertsilicon tubing was used in order to prevent stresspotentially exerted by plasticisers. The constant per-turbation effectively resulted in mixing that preventedthe water from becoming dysoxic. A lid covered eachbioreactor in order to minimise eva pora tion and sta-bilise the pCO2 in the headspace filled with the gasmixture. Humidifying the gas flow in gas-wash bot-tles filled with de-ionised water further reduced riskof evaporation, which was found to be negligible at0.2 ± 0.1% d−1 on average (n = 60).

The 72 h exposure period in potentially stressfulconditions inherent to laboratory experiments orcrowded aquaria (Osinga et al. 1999) is well withinthe tolerance limits of Cliona orientalis. We tested ourexperimental setup in an additional 5 d experimentalrun at ambient temperature (25°C) and present-daypCO2. Daily carbonate system assessments indicatedthat all sponges continuously bioeroded and survivedwith few if any stress symptoms. In addition, after theseries of pCO2 × temperature treatments, all surviv-ing sponges were transferred back into an outdoorflow-through system and fully recovered.

Carbonate system parameters

As a basis for an accurate assessment of the amountof CaCO3 dissolved by Cliona orientalis during theexperiment, the seawater carbonate system and thevariables characterising or controlling this systemwere closely monitored (equilibria between aqueouscarbon dioxide CO2aq, bicarbonate HCO3

−, and car-bonate ions CO3

2−; see Zeebe & Wolf-Gladrow 2001for a review). This comprised measurements of thedissolved inorganic carbon (DIC), total alkalinity (TA),pH, temperature, salinity, and dissolved inorganicnutrients. Based on these variables, the remainingcarbonate system parameters, as well as the dissolu-tion of carbonate, were quantified (see Table 1).

During the experiment, temperature was re -corded in 5 min intervals with Starmon Mini high-resolution loggers (Star Oddi; accuracy: ±0.02°C).Water samples were taken and sterile-filtered with0.2 µm PES filters at the start and end of the ex -periment. Samples for DIC and TA were treatedwith 0.02 vol % saturated HgCl2 solution to arrestbiological activity, whereas samples for nutrientanalyses were left untreated, but stored refriger-ated and in the dark. Before water sampling, salin-ity was measured on the PSS scale using a Sev-enGo DUO meter (Mettler-Toledo) equipped withan InLab 738-ISM conduc tivity probe calibrateddaily with 12.88 mS cm−1 NIST-certified buffersolution. The relative order and timing of all pro-cedures was kept constant over the course of the ex -periments to avoid natural diurnal variation (Wisshaket al. 2012).

114

Fig. 2. Setup of experimental system situated in a constant temperature room at Orpheus Island Research Station with individ-ual semi-closed bioreactors (4 replicates per treatment level), carrying 8 sponge-bearing cores each, and 1 bioreactor with sea-water only for monitoring purposes, all situated in a larger tank to maintain the respective treatment temperature. The moder-ately and strongly elevated pCO2 levels were adjusted by perturbation with specifically mixed gases; for the present-dayscenario, compressed air was used. Pre-filtered seawater for all treatments was taken from an aerated 600 l reservoir tank

connected to a pump and UV unit

Wisshak et al.: Global change impacts on sponge bioerosion

Prior to chemical analyses, pH of the water sampleswas quantified on the total scale using a flat-membrane glass electrode (6.0256.100; DeutscheMetrohm) calibrated against seawater standards ofknown pH to match ionic strength between samplesand buffer solutions. Nitrate, nitrite, and phosphatewere evaluated photometrically (U-2000; Hitachi)according to standard methods described by Hansen& Koroleff (1999) with precision levels of ±0.5, ±0.02,and ±0.05 µmol l−1, respectively; ammonium wasquantified fluorometrically (SFM 25; Kontron In -struments) according to Holmes et al. (1998) to±0.08 µmol l−1. TA was determined in duplicate, usingpotentiometric open-cell titration following Dicksonet al. (2003). Seawater was weighed (1416B MP8-1;Sartorius) and titrated with 0.005 N HCl in an auto-matic titrator (Titrando 808; Deutsche Metrohm); theaverage precision between duplicate water sampleswas ≤4 µmol kg−1. DIC was measured photochemi-cally following Stoll et al. (2001) using an automatedsegmented flow analyser (QuAAtro; Bran+Luebbe)equipped with an autosampler (±5 μmol kg−1 pre -cision). Both TA and DIC were calibrated with certi-fied seawater reference material (CRM standardssupplied by Andrew Dickson, Scripps Institution ofOceanography).

The carbonate system was computed from theobtained temperature, salinity, TA, and pH usingthe CO2SYS program (EXCEL macro v. 2.1; Pierrotet al. 2006) with dissociation constants for carbonicacid from Mehrbach et al. (1973) as refit byDickson & Millero (1987) and the KSO4 dissociationconstant from Dickson (1990). TA and pH werechosen for the carbonate system calculations inorder to avoid inconsistencies in carbonate chem-istry potentially affected from calculations by DICand TA (Hoppe et al. 2012). Moreover, as a controlof the whole procedure, the measured pH valueswere compared with calculated pH values (via DICand TA) and found to differ by only 1.3 ± 1.6% onaverage (n = 108).

Assessment of bioerosion rates

Carbonate dissolution (mg bioreactor–1) was com-puted by adapting the alkalinity anomaly tech-nique (Smith & Key 1975, Chisholm & Gattuso 1991)from the change in total alkalinity ΔTA during the72 h of incubation, involving a correction for nitro-gen nutrients and phosphate according to Jacques &Pilson (1980), and using the TA definition of Dicksonet al. (2003) (Eq. 1).

ΔCaCO3 = 0.5 × [ΔTA + ΔPO4 − ΔNH4 + Δ(NO3+ NO2)]× 100 × Vsw × ρsw/103 (1)

The factor 0.5 reflects the 2:1 relationship betweenthe increase of TA during carbonate dissolution, themultiplier 100 is the molecular mass of CaCO3, Vsw

the seawater volume per bioreactor (l), ρSW the density of seawater (kg l−1) as a function of tem -perature and salinity calculated with R using the pack -age ‘SeaCarb’ version 2.4 (www.cran.r-project.org/package=seacarb), and the divider 103 converts tothe weight unit milligrams.

Evaluation of bioerosion in the 31°C treatment wascomplicated by cases of sponge mortality in 5 out ofthe 12 bioreactors. The affected sponge tissue turnedblack, fouling the water and making it more viscousand impossible to filter, thus disabling accuratemeasurements for carbonate system calculations—therefore data from these bioreactors were excludedfrom all analyses except for a health assessment.The remaining 7 bioreactors in the 31°C treatmentyielded high nutrient concentrations at the end ofthe experiment, with ammonium levels above thethresh old for accurate determination (dilution factor>8). In order to allow adequate correction for nutri-ents, a conservative estimation of 100 µmol l−1 wasapplied, i.e. a level just above the highest values ofammonium with measurements within the thresholdof accurate determination.

Dissolved CaCO3 was then converted to chemicalbioerosion rates as kg m−2 yr−1 by relating the valueto the combined sponge tissue surface area of the 8cores in each bioreactor. Core-sampling generatednew surface with feeding apertures, i.e. the spongewas not only supported by the upper circle of originalsurface, but also by a ring of healed surface aroundthe sides of the cores. As bioerosion underneath thisring-shaped surface area acted on skeleton alsounderneath the upper surface area, we included onlyhalf of this surface into the reference unit in order notto underestimate bioerosion rates. Ring surface areawas calculated using the mean tissue penetrationdepth of 1.32 ± 0.14 cm, which was determined frommin/max measurements in 128 sponge-bearing corestaken from 10 different Cliona orientalis colonies inLittle Pioneer Bay.

Statistical analyses

All statistical calculations used means per bioreac-tor, which represented the replicate level. Despiterunning the temperature treatments in a temporal se-

115

Aquat Biol 19: 111–127, 2013

quence, they were regarded as independent, be causeeach time the experiment was adjusted to a new tem-perature and new sponge cores were used. Mortalityoccurred in the 31°C stress treatment, which resultedin the removal of 5 replicates (bio reactors) in the re-spective datasets relating to photosynthesis para -meters and bioerosion, but ‘sponge health’ remainedfully replicated, because ‘death’ was part of the healthassessment and a result. Increased variation of re-sponse values in the extreme treatments also meantthat heteroscedasticity was common, and some otherassumptions such as normal distribution were occa-sionally violated as well, which could not always beremoved by transfor mation. Simple non-parametricKruskal-Wallis tests allowed for neither the recogni-tion of interaction effects nor for post hoc analysis attreatment level. Therefore, we looked for general dif-ferences in 5 separate, non-parametric permutationalmultivariate analysis of variance (PERMANOVA)models calculated in the software Primer 6 andPERMANOVA (Anderson 2005) based on resem-blance matrices with Euclidean distance, with 9999permutations of residuals under a reduced model andwith Type III sums of squares (Anderson et al. 2008;models for ‘health score’, ‘Fv/Fm’, ‘F0’, ‘bioerosionrate’, and ‘ΔNH4’). The experimental factors were en-tered as fixed effects (pCO2 with 3 levels: present-day, moderately elevated, and strongly elevated;temperature with 4 levels: 22, 25, 28, and 31°C). Sig-nificant results were followed up with PERMANOVApairwise comparisons. Evaluations of ‘health score’,‘Fv/Fm’, and ‘F0’ partly depended on chlorophyll con-centrations in the surface layer of the sponge. Theyslightly differed between sponge individuals and alsowere slightly affected by the acclimatisation to con-stant dark (see Fig. 4 for differences comparing ‘be-fore’ with ‘after’ values employing Wilcoxon signedrank tests performed in SPSS 19, IBM). Therefore,these parameters were statistically tested in PERM-ANOVA as % of the originally measured value. Wethen checked relationships between treatments andresponse values in multiple linear regressions inSPSS. While this was used to answer the centralcause-effect query, we also displayed re sults of addi-tional simple linear regressions to obtain more detailed information on causes and responses and to provide equations for individual settings, e.g. formodelling purposes. The linear regressions may alsobe useful, because the relationship between ‘bioero-sion rate’ and temperature remained unclear, and atthe same time a weak interaction effect was foundbetween temperature and pCO2. Therefore, for ‘bio-erosion rate’, it is more accurate to regard the results

for subsets of data than for the multiple linear regres-sion. Slope equations resulting from regressions ledto predictions for bioerosion under future conditions(SigmaPlot 12; Systat).

RESULTS

General physiological responses

Physiological responses not directly related to bio-erosion were monitored for Cliona orientalis: healthof the sponge cores and their photosymbionts byvisual assessment and by PAM, mortality, recovery,and quantification of dissolved nutrients in the cul-ture water (Figs. 3 & 4, Table 1). An overall slight pal-ing of sponge cores was observed after 1 d in dark-ness for acclimatisation (cf. Fig. 3a & 3d). Whilesponges in the 22°C treatment retained their healthyappearance, paling increased with higher tempera-tures, and F0 decreased (Fig. 3d), suggesting thatgradually more zooxanthellae were moved awayfrom the surface and deeper into the sponge body. Anumber of dead or partially dead cores occurredin the 31°C stress temperature treatment (Fig. 3c),which was the only situation in which photosyntheticefficiency was also slightly lowered (Fig. 4f). Visualhealth score and photosynthetic parameters decreasedbetween ‘before’ and ‘after’ the experiment (Fig. 3d−f).However, these responses only differed with andwere only correlated to temperature increase, not toincreased pCO2 (Table 2; and Tables S1 & S2 in theSupplement at www.int-res.com/articles/suppl/ b019p111_supp.pdf). Thus no negative effect could beattributed to elevated pCO2, but it may have reducedthe temperature stress, as evidenced by our bargraphs (Fig. 4c,f,i) and possibly supported by interac-tion effects near the significance threshold for Fv/Fm

and F0 (Table 2). All cores fully recovered within aweek after being placed back into the outdoor flow-through tanks (Fig. 3b), except for those that werealready dead or nearly dead. Cores with smallerblack patches lost the affected tissue but recovered.

Initial levels of all examined nutrients in the fil-tered culture water were very low at 0.06 ± 0.11 µmoll−1 NO3, 0.02 ± 0.02 µmol l−1 NO2, 0.04 ± 0.04 µmol l−1

PO4, and 0.48 ± 0.85 µmol l−1 NH4 (all replicates; n =48). At the end of the experiment, the levels fornitrate, nitrite, and phosphate had increased but re -mained at a low level, with mean values per treat-ment <2.60 µmol l−1 (Table 1). In contrast, ammoniumlevels increased and varied considerably. This in -crease (ΔNH4 = 10.95 ± 15.78 µmol l−1 at 22°C, 28.91

116

Wisshak et al.: Global change impacts on sponge bioerosion

± 21.70 µmol l−1 at 25°C, 44.09 ± 32.97 µmol l−1 at28°C, and a critical level of 98.73 ± 15.78 µmol l−1 at31°C) showed significant correlation with tempera-ture but not with pCO2 (Table 2).

Bioerosion rates

Bioerosion by Cliona orientalis differed signifi-cantly with pCO2 and temperature, and the factorsinteracted (Figs. 4j & 5, Table 2). Interaction was most

likely caused by the fact that pairwise comparisonsbetween pCO2 levels were always significant, exceptat 31°C, where omission of replicates resulted in theloss of support and the failure of finding significances(Table S1). This is expected to be a Type II error ofnot finding a difference when there should be one,because the plotted data still display the same trendas for the other 3 temperature levels (cf. Fig. 5d and5a−c). However, in contrast to a clear linear anddirectly proportional relationship with pCO2, and con -sequently a negative correlation with pH (Fig. 5a−d,

117

Fig. 3. Monitoring of Cliona orientalis health in symbiosis with zooxanthellae, ‘before’ (= after acclimatisation) and ‘after’ theexperiment in continuous dark. Only representative figures for different temperature treatments are shown; different pCO2

did not change the appearance within temperature levels. One replicate set of 8 sponge cores each (a) ‘before’ the experiment,(b) ‘after’ the experiment and a recovery period of several days in ambient temperature and flow-through conditions, and (c) aset of dead cores from the 31°C treatment. Appearance of cores at each temperature level (d) ‘before’ and ‘after’ the experi-ment, and corresponding pulse-amplitude modulated fluorometry (PAM) images illustrating (e) photosynthetic efficiency

Fv/Fm and (f) the proxy for chlorophyll concentration F0

Aquat Biol 19: 111–127, 2013118

Fig. 4. Means of experimental response values (a,d,g) ‘before’ (= after acclimatisation period) and (b,e,h) ‘after’ the experi-ment, and as (c,f,i) percent of ‘before’, complemented by (j) the respective rates of chemical bioerosion. Error bars are SD. Dis-played significances compare ‘before’ and ‘after’ values using a non-parametric Wilcoxon signed-rank test (partial tests sepa-rating out the effects of pCO2 and temperature (T) were conducted for ambient conditions of the respective other treatmentfactor). Significant values in bold. Five replicates for all ‘after’ and percentage datasets except for the health score were

omitted due to mortalities in the 31°C treatment series

Wisshak et al.: Global change impacts on sponge bioerosion 119

Temp. Variable Present-day Moderately elevated Strongly elevatedlevel pCO2 pCO2 pCO2

22°C Temperature (°C) 21.8 ± 0.2 22.1 ± 0.2 22.1 ± 0.2Salinity (PSS) 34.2 ± 0.0 34.3 ± 0.0 34.3 ± 0.0pH (total scale) 8.00 ± 0.01 7.80 ± 0.01 7.52 ± 0.01TA (µmol kg−1) 2305.3 ± 12.6 2384.9 ± 10.2 2607.3 ± 33.7ΔTA (µmol kg−1) 112.5 ± 23.2 266.8 ± 19.2 711.3 ± 67.1DIC (µmol kg−1) 2061.2 ±16.8 2227.8 ± 13.3 2553.2 ± 35.0pCO2 (µatm) 455.0 ± 19.6 804.6 ± 27.3 1782.2 ± 49.1HCO3 (µmol kg−1) 1869.4 ± 19.1 2076.3 ± 14.3 2420.5 ± 33.7CO3 (µmol kg−1) 177.7 ± 3.8 126.8 ± 2.3 78.1 ± 1.5ΩAr 2.80 ± 0.06 2.00 ± 0.04 1.23 ± 0.02ΩCa 4.29 ± 0.09 3.06 ± 0.06 1.88 ± 0.04ΔNO3 (µmol l−1) 0.49 ± 1.14 1.26 ± 0.99 0.41 ± 0.83ΔNO2 (µmol l−1) 0.21 ± 0.46 0.74 ± 0.47 0.18 ± 0.31ΔNH4 (µmol l−1) 14.31 ± 19.33 16.67 ± 19.10 1.87 ± 2.41ΔPO4 (µmol l−1) 0.86 ± 0.43 1.64 ± 0.19 0.86 ± 0.59Chemical bioerosion rate (kg m−2 yr−1) 0.15 ±0.04 0.37 ± 0.05 1.04 ± 0.10

25°C Temperature (°C) 23.9 ± 0.2 24.5 ± 0.2 24.5 ± 0.2Salinity (PSS) 34.3 ± 0.1 34.3 ± 0.0 34.3 ± 0.0pH (total scale) 8.04 ± 0.00 7.84 ± 0.01 7.57 ± 0.01TA (µmol kg−1) 2317.9 ± 8.2 2414.1 ± 9.8 2614.0 ± 25.3ΔTA (µmol kg−1) 178.0 ± 13.4 368.0 ± 22.9 766.5 ± 41.6DIC (µmol kg−1) 2031.8 ± 6.3 2221.5 ± 12.9 2526.3 ± 26.2pCO2 (µatm) 418.2 ± 2.6 742.2 ± 21.7 1594.7 ± 53.5HCO3 (µmol kg−1) 1814.0 ± 5.0 2050.5 ± 14.3 2384.7 ± 25.7CO3 (µmol kg−1) 205.7 ± 1.8 150.0 ± 2.5 96.3 ± 3.1ΩAr 3.27 ± 0.03 2.39 ± 0.04 1.54 ± 0.05ΩCa 4.98 ± 0.04 3.63 ± 0.06 2.33 ± 0.08ΔNO3 (µmol l−1) 1.16 ± 0.36 0.75 ± 0.59 1.23 ± 0.55ΔNO2 (µmol l−1) 0.90 ± 0.14 0.94 ± 0.29 0.97 ± 0.13ΔNH4 (µmol l−1) 35.88 ± 31.10 20.96 ± 11.47 29.90 ± 21.87ΔPO4 (µmol l−1) 2.25 ± 0.85 2.58 ± 0.95 1.99 ± 0.97Chemical bioerosion rate (kg m−2 yr−1) 0.21 ± 0.04 0.51 ± 0.04 1.09 ± 0.04

28°C Temperature (°C) 27.8 ± 0.3 27.9 ± 0.3 27.9 ± 0.3Salinity (PSS) 34.3 ± 0.0 34.3 ± 0.0 34.3 ± 0.0pH (total scale) 8.03 ± 0.02 7.81 ± 0.02 7.55 ± 0.02TA (µmol kg−1) 2283.2 ± 10.0 2359.0 ± 20.7 2583.5 ± 20.2ΔTA (µmol kg−1) 90.7 ± 19.4 238.5 ± 41.9 676.2 ± 23.7DIC (µmol kg−1) 1973.4 ± 21.8 2157.2 ± 26.5 2487.5 ± 14.0pCO2 (µatm) 418.9 ± 28.8 773.8 ± 42.9 1676.8 ± 67.1HCO3 (µmol kg−1) 1741.4 ± 28.8 1981.8 ± 29.0 2342.9 ± 11.6CO3 (µmol kg−1) 221.0 ± 8.0 155.2 ± 4.6 100.9 ± 4.6ΩAr 3.58 ± 0.13 2.52 ± 0.08 1.64 ± 0.08ΩCa 5.39 ± 0.19 3.78 ± 0.11 2.46 ± 0.11ΔNO3 (µmol l−1) 1.06 ± 0.77 2.06 ± 0.83 2.07 ± 0.75ΔNO2 (µmol l−1) 1.51 ± 0.12 1.99 ± 0.75 1.87 ± 0.49ΔNH4 (µmol l−1) 31.98 ± 30.48 70.90 ± 5.25 29.39 ± 39.83ΔPO4 (µmol l−1) 1.61 ± 1.24 1.76 ± 0.63 0.97 ± 0.63Chemical bioerosion rate (kg m−2 yr−1) 0.09 ± 0.02 0.25 ± 0.06 0.95 ± 0.08

31°Ca Temperature (°C) 30.9 ± 0.1 30.8 ± 0.1 30.8 ± 0.1Salinity (PSS) 34.8 ± 0.0 34.9 ± 0.0 34.9 ± 0.0pH (total scale) 8.07 ± 0.01 7.92 ± 0.01 7.56 ± 0.01TA (µmol kg−1) 2350.2 ± 16.9 2403.7 ± 13.4 2705.9 ± 86.7ΔTA (µmol kg−1) 180.7 ± 23.4 279.9 ± 33.2 900.7 ± 171.1DIC (µmol kg−1) 1974.6 ± 24.1 2120.4 ± 8.9 2581.5 ± 86.5pCO2 (µatm) 380.0 ± 21.4 598.6 ± 8.8 1710.8 ± 65.2HCO3 (µmol kg−1) 1699.2 ± 27.6 1897.3 ± 6.6 2417.4 ± 82.5CO3 (µmol kg−1) 266.0 ± 4.1 208.3 ± 3.8 122.1 ± 2.4ΩAr 4.36 ± 0.07 3.41 ± 0.06 2.00 ± 0.04ΩCa 6.50 ± 0.10 5.08 ± 0.09 2.98 ± 0.06ΔNO3 (µmol l−1) 0.90 ± 0.02 1.69 ± 1.00 0.15 ± 0.21ΔNO2 (µmol l−1) 2.03 ± 0.41 2.15 ± 0.39 1.92 ± 0.11ΔNH4 (µmol l−1)b 98.94 ± 0.82 97.93 ± 2.36 99.73 ± 0.38ΔPO4 (µmol l−1) 1.76 ± 0.90 2.29 ± 1.61 2.43 ± 1.82Chemical bioerosion rate (kg m−2 yr−1) 0.13 ± 0.03 0.28 ± 0.05 1.18 ± 0.25

a5 replicates excluded due to mortality in the 31°C treatmentbEnd concentrations exceeding upper limit of measurement range were set to 100 µmol l−1

Table 1. Experimental settings related to the carbonate system. List of measured (temperature, salinity, pH, mean and Δ totalalkalinity [TA]) and computed carbonate system parameters (dissolved inorganic carbon [DIC], carbon dioxide partial pres-sure [pCO2], hydrogen carbonate [HCO3] and carbonate ion [CO3] molality, saturation states for aragonite [ΩAr] and calcite[ΩCa]), nutrients (nitrate [NO3], nitrite [NO2], ammonium [NH4], phosphate [PO4]), and rates of chemical sponge bioerosion.Salinity, pH, and TA were measured in duplicate. All values are given as mean ± SD among the replicates after 3 d of exposure

Aquat Biol 19: 111–127, 2013

Tables 2, S1 & S2), no clear relationship with tem -perature could be demonstrated, also evidenced byonly few and weakly significant differences foundwhen testing temperature levels pairwise againsteach other (Fig. 4j, Tables 2, S1 & S2). The weakerand somewhat inconclusive temperature effect wasconfirmed by the multiple linear re gressions yieldinga clear significance for the factor pCO2 and a neg -ligible contribution of temperature (Table S2). Atany given pCO2 level, bioerosion rates were highestat 25°C (approx. ambient winter temperature), andslightly de creased towards colder as well as to wardswarmer tem peratures (Fig. 4j). Temperature effectson bioerosion rates were most pronounced between25 and 28°C, whereas for 22 and 31°C, intermediaterates were determined (Table S1). The well-definedslopes of the linear regressions for bioerosion at dif-ferent levels of pCO2 were nearly identical for 22, 25,28, and 31°C, which again shows the low impact of

temperature (Fig. 5a−d). Based on the very well sup-ported results of the multiple linear regression, andthe fact that the overall equation is very clearlydriven by pCO2, rather than temperature (Fig. 5e,Table S2), equations for chemical sponge bioerosionrates with respect to pCO2 (Eq. 2) and to pH (Eq. 3)were derived from this, complemented by a regres-sion expressing the % increase relative to the pres-ent-day pCO2 (Eq. 4):

Chemical bioerosion rate (kg m−2 yr−1) = 0.0007 pCO2 (µatm) − 0.1547 (2)

Chemical bioerosion rate (kg m−2 yr−1) = −1.8665 pH + 15.082 (3)

Chemical bioerosion (% of present-day rate) = 0.5989 pCO2 (µatm) − 130.79 (4)

The latter equation was applied for plotting thepercent increase in chemical sponge bioerosion for 3

of the Special Report on Emissions Sce-narios (SRES) as illustrated in Fig. 6.Keeping methodological differences inmind (see ‘Discussion’), a comparison ofthe regression for total Cliona orientalis bioerosion (data from Wisshak et al. 2012)with the chemical bioerosion rates re -ported herein allows a tentative deduc-tion of the ratio between chemical andmechanical bioerosion and its change withincreasing pCO2 (Fig. 7). While underpresent-day conditions apparently only9.2% of the substrate was removed bychemical dissolution, this fraction al mosttripled to 26.5% at 1500 µatm pCO2.

DISCUSSION

General conclusions relating to theexperimental conditions

We were able to show that, despite con-ducting the experiment in continuousdarkness, the photosymbiotic spongeCliona orientalis was not impaired for theshort duration of the experiment. Zoo-xanthellae were shifted deeper into thesponge in darkness, as was expected(Schönberg & Suwa 2007), but this pro-cess was not uniform across treatmenttemperatures. Monitoring its visual ap -pearance and the photo synthetic para -meters Fv/Fm and F0 demonstrated

120

Source of df SS MS Pseudo-F p Unique variation (perm) perm

Health score (% of ‘before’ values)pCO2 2 163.9 82.0 0.4199 0.6606 9953T 3 20272.0 6757.3 34.6160 0.0001 9957pCO2 × T 6 1203.3 200.6 1.0273 0.4298 9948Residual 36 7027.5 195.2Total 47 28667.0

Fv/Fm (% of ‘before’ values)a

pCO2 2 4.0 2.0 1.3826 0.2655 9959T 3 593.4 197.8 136.1600 0.0001 9949pCO2 × T 6 19.6 3.3 2.2476 0.0621 9948Residual 31 45.0 1.5Total 42 641.0

F0 (% of ‘before’ values)a

pCO2 2 36.5 18.2 1.3 0.2995 9952T 3 2390.3 796.8 55.8 0.0001 9935pCO2 × T 6 260.8 43.5 3.0 0.0191 9930Residual 31 442.7 14.3Total 42 3201.6

Chemical bioerosion rate (kg m−2 yr−1)a

pCO2 2 6.0 3.0 597.8 0.0001 9961T 3 0.2 0.1 11.7 0.0001 9951pCO2 × T 6 0.1 0.0 3.2 0.0132 9949Residual 31 0.2 0.0Total 42 6.7

ΔNH4 (µmol l−1)a

pCO2 2 1129.8 564.9 1.2 0.3131 9945T 3 27577.0 9192.5 19.5 0.0001 9947pCO2 × T 6 3930.1 655.0 1.4 0.2456 9955Residual 31 14604.0 471.1Total 42 48357.0a5 replicates excluded due to mortality in the 31°C treatment

Table 2. PERMANOVAs with Type III sums of squares and 9999 permuta-tions of residuals for the dependent variables health score, photosyntheticefficiency (Fv/Fm), chlorophyll concentration (F0), chemical bioerosion rate,and ΔNH4, with fixed effects, 3 levels for carbon dioxide partial pressure

(pCO2), and 4 for temperature (T) . Significant values in bold

Wisshak et al.: Global change impacts on sponge bioerosion 121

Fig. 5. Rates of chemical bioerosion per replicate bioreactor asa function of pCO2 levels (present-day, moderately elevated,strongly elevated), and for the various temperature settings,with (a) ~22°C (colder than ambient), (b) ~25°C (ambient), (c)~28°C (warmer than ambient), (d) ~31°C (stress temperature;representing only replicates without mortality), and (e) alltreatments including the percent change relative to the pres-ent-day rates (blanks omitted). Linear regressions are plottedwith 95% confidence and prediction bands and are highly

significant for all treatments

Aquat Biol 19: 111–127, 2013

that while photosynthetic efficiency was temporarilyreduced in cores in the 31°C treatment (Tables 1, 2& S1), and health appearance and chlorophyll con-centrations decreased with increasing temperature(Figs. 3 & 4, Tables 1, 2 & S1), all surviving coresquickly recovered to their original appearance underflow-through/ambient con ditions after the experi-

ment (Fig. 3b). While strongly increased temperatureapparently had a negative effect on the sponge’shealth and on the photosynthesis of its symbionts,pCO2 did not (Tables 2 & S1), or might even have hada slightly ameliorating effect on the temperaturestress, a hypothesis which may be supported byinteraction effects between temperature and pCO2

found near the significance level (Fig. 4c, Table 2).The effects of pCO2 on bioerosion can thus be identi-fied as direct effects and were not caused by indirectimpacts through lowered fitness.

The experimentally determined carbonate dissolu-tion can be assigned with confidence to chemical bio-erosion by Cliona orientalis. In all 12 settings of thepCO2 × temperature treatment matrix, the cultureseawater was saturated with respect to both arago-nite and calcite (Ω > 1; Table 1), even under condi-tions where the lowest saturation state would beexpected, i.e. the 22°C treatment with strongly ele-vated pCO2. In order to maintain a stable pH and saturation regime without a diurnal rhythmicity in -duced by CO2 uptake during photosynthesis of thephotosymbionts and by increasing temperatures during the illumination phase, such as recorded byWisshak et al. (2012), the experiment was run in con-stant darkness. Therefore, our bioerosion values maybe underestimations, because zooxanthellate clion-aids receive supplementing nutrients from their symbionts, and over longer periods erode more inlight than in shade (Schönberg 2006 and referencestherein, Weisz et al. 2010). However, over a shortperiod of only few days, this would not significantlyaffect bioerosion, as filter-feeding was not impaired,and clionaids are also known to assimilate dissolvednutrients from the water column (Maldonado et al.2012). Moreover, clionaids may switch their symbi-otic relationships depending on circumstances anddigest zooxanthellae (Rosell 1993, Hill & Hill 2012).We would expect a lag period before physiologicalfunctions would be affected by the missing additionfrom the symbionts. A potential contribution of non-sponge endoliths to the recorded bioerosion in thecoral cores that would have led to overestimatedsponge bioerosion rates was assumed negligiblebased on the observation that substrate occupied byC. orientalis is not usually infested by other macro-borers. Moreover, the overlying sponge tissue appar-ently prevents establishment of a significant micro -endolith community (coral matrix and potential micro -endoliths = mass of organics in coral substrate madeup less than 1%; Fang et al. 2013). Important addi-tional support for reporting unbiased sponge bio -erosion is provided by the fact that control cores in a

122

Fig. 6. Projected percent increase in chemical sponge bio-erosion relative to the present-day level, calculated for theBERN-CC model based on the Special Report on Emissions

Scenarios (SRES) A2 (red), A1B (blue), and B1 (green)

Fig. 7. Chemical bioerosion rates from the present study’s25°C treatment compared to the 25°C total bioerosion ratesreported by Wisshak et al. (2012). The significant increase intotal bioerosion rates is to a large extent governed by the increase in chemical bioerosion, while mechanical bioero-sion increases only slightly. The fraction of chemical versustotal bioerosion increases from 1/10 at 390 µatm to 1/5 at

1000 µatm and 1/4 at 1500 µatm

Wisshak et al.: Global change impacts on sponge bioerosion

previous experiment did not lose weight during 10 dof exposure (Wisshak et al. 2012). While bacterialbioerosion in the 31°C treatment cannot be dis-counted, tissue death only occurred late in the exper-iment, thus only allowing a short period for bias, andapart from that, bioreactors with dead sponges wereexcluded from the calculations for bioerosion. Wefurthermore ruled out unrelated shifts in the carbon-ate system of the culture seawater by monitoringbioreactor blanks that did not contain any sponge orcoral material, none of which exhibited a relevantΔTA over the course of the experiment (Fig. 5a−d).Hence, the positive ΔTA in the bioreactors containingsponge cores is interpreted to be caused solely bychemical sponge bioerosion.

Influence of temperature and pCO2 on spongebioerosion

Temperature effects on Cliona orientalis bioero-sion rates did not follow a clear linear trend in ourexperiment (Fig. 4j, Table S2) and may in part alsoreflect effects of the ammonium building up. Thehighest bioerosion rates were reached at the localambient temperature of 25°C, decreasing towardscolder as well as warmer tem peratures. Overall, theinfluence of temperature on the bioerosion capacityof C. orientalis appears to be very low to negligibledespite other physiological responses caused by heat(Fig. 4, Tables 1, 2, S1 & S2), particularly with respectto ammonium, that can be interpreted as a sign ofexcretion and/or disintegrative processes (Wright1995, Francis-Floyd et al. 2009). The ob served signif-icant positive correlation of ammonium with temper-ature (Table 2) likely suggests increasing heat stress,with resulting ammonium and/or ammonia poten-tially acting as an additional indirect stressor (Bower& Bidwell 1978, Osinga et al. 1999). In contrast totemperature, pCO2 does not correlate with theobserved ammonium levels (Table 2). However, highpH favours the formation of the more toxic ammonia(Osinga et al. 1999), which may partly explain theslightly more negative res ponse in the low-pCO2

treatments (Fig. 4). Little is known about the rate ofammonium/ammonia gen eration and tolerance insponges (Osinga et al. 1999). According to Sipkemaet al. (2004), isolated cells of the Medi terraneandemosponge Suberites domun cula (Olivi, 1792) werenot affected by very high ammonium treatments (upto 25 000 µmol l−1; compared to our maximum of 100µmol l−1), but were clearly negatively affected byheat (12 to 28°C). Moreover, the sponge samples

originated from a coastal reef that is subject to largefluctuations of seasonal and disturbance-relatedwater quality (e.g. Fabricius & Dommisse 2000); how-ever, they are very successful at this site (Schönberg& Ortiz 2009). We thus regard the overall effect ofammonium as smaller than the heat treatment. Whilethe increase of ammonium levels across the tempera-ture treatments was significant, the concentrationsalone may not have been high enough to cause theobserved mortality, but the strongly elevated ammo-nium in the 31°C treatment may well be a result oftissue death towards the end of the experiment.

Sponge cores that survived the stress temperatureyielded comparatively high bioerosion rates. Never-theless, critically high sea surface temperatures mayharm bioeroders to a similar degree as calcifiers andmay trigger diseases, reduction in symbiont viability,and mortality in both organism guilds (Osinga et al.1999, Schönberg et al. 2008).

Temperature having little influence on sponge bio-erosion rates was surprising and could possibly beexplained by 3 partly antagonistic processes:

(1) Acceleration of biochemical processes withincreasing temperature may lead to increased ratesof chemical bioerosion within metabolic tolerancelimits (van’t Hoff equation and its derivations: 2 to 4×reaction speed with every 10°C increase; Q10 factor>1; Atkins & De Paula 2009).

(2) Rising temperatures will reduce the physico-chemical dissolution capacity of CaCO3 in seawater(Mehrbach et al. 1973) and may thus decrease bio-erosion capabilities.

(3) Metabolic processes inherent to Cliona orien-talis may function best at local ambient temperaturethe sponge is adapted to.

Options (1) and (2) may cancel each other out tosome degree, but will only function within the limitsof (3). The latter would also explain the observationof most pronounced bioerosion recorded at ambientconditions, because re-adaption to colder or warmertemperatures may cost energy and thus lead to atemporary loss in bioerosion efficiency as a trade-off.

Our observation of temperature having little orpossibly a non-linear effect on bioerosion rates wasmost recently confirmed for a related species, Clionacf. celata Grant, 1826 (explants on scallop shells;Duckworth & Peterson 2013). The latter authors foundno significant relationship between shell weight lossand temperature, using 26°C as an ambient settingand 31°C for the year 2100.

Cliona orientalis chemical bioerosion rates consis-tently and significantly increased with increasingpCO2, and the present observations are thus in good

123

Aquat Biol 19: 111–127, 2013

accordance with earlier findings on total sponge bio-erosion of the same species (Wisshak et al. 2012), aswell as that of the related C. cf. celata (Duckworth &Peterson 2013; acid-regulated experiment with bio-erosion rates doubling from pH 8.1 to 7.8). Thisclearly confirms our initial hypothesis that spongebioerosion can be predicted to escalate with pro-gressing OA. Applying Eq. (4), the increase of chem-ical bioerosion by C. orientalis converts to a 268.3%escalation by the end of this century, following theBERN-CC model based on the Special Report onEmissions Scenarios (SRES) A2 emission scenario(2100 pCO2 = 836 µatm; IPCC 2001). The most opti-mistic B1 model (pCO2 = 540 µatm) would lead to a91.0% increase in C. orientalis chemical bioerosion,and the moderate A1B model (pCO2 = 703 µatm) toan elevation of 185.2% (Fig. 6). The observed rise inchemical bioerosion is one order in magnitude strongerthan for total bioerosion of C. orientalis as predictedby Wisshak et al. (2012; compare to Fig. 5b therein).Pre dictions based on the Representative Concentra-tion Pathways (RCPs; Meinshausen et al. 2011)—aset of 4 scenarios for use in climate modelling—result in similar figures of 86.5, 164.1, and 320.6%increase in C. orientalis chemical bioerosion forthe RCP 4.5 (2100 pCO2 = 538 µatm), 6 (pCO2 =670 µatm), and 8.5 (pCO2 = 936 µatm) scenarios, re -spectively. Hence, doubling to quadrupling of chem-ical bioerosion by C. orientalis can be expected bythe year 2100, depending on which emission scenariois considered. At the same time, according to theIPCC (2001), surface air temperatures are expectedto rise by 1.98°C (SRES B1 model) to 3.79°C (A2model), which may push many tropical coral reefsinto a temperature range where bioeroding spongesmay be seasonally exposed to heat stress and possi-bly to mortality, thus limiting the range of the pCO2

effect on reef bioerosion.Sponge bioerosion is therefore much more clearly

and strongly influenced by pCO2 than by tempera-ture, and follows a directly proportional, linear trendfor pCO2, while the response to temperature and/orammonium is more complicated. Temperature maycause a threshold response with respect to the sponge’sphysiology, and bioerosion ceases when other physi-ological functions are impaired. In contrast, the rela-tionship between pCO2 and sponge bioerosion mayact along 2 pathways: the acceleration of chemicalbioerosion by lowering the pH gradient that needs tobe overcome in the related biochemical reaction(Wisshak et al. 2012), and the enhancement of thesponges’ energy budget by translocation of carbonstemming from the photosymbionts (Weisz et al.

2010), a process that may be facilitated by OA, notinhibiting the sponge’s physiology.

Considering the important role of bioerodingsponges for reef carbonate budgets (e.g. Acker &Risk 1985, Schönberg 2002b, Nava & Carballo 2008,Perry et al. 2008), observations that their abundancesincreased after disturbance and stress (e.g. Rützler2002, Ward-Paige et al. 2005, Schönberg & Ortiz2009), the fact that other organisms using chemicalbioerosion exhibited similar trends in climate changestudies (Tribollet et al. 2006, 2009, Reyes-Nivia etal. 2013), and that advanced internal bioerosion facilitates external bioerosion by grazers (e.g. Tri -bollet et al. 2011), we fully expect that total bioero-sion will increase in future until the physiologicallimits of the involved organisms are reached. How-ever, while short-term aquarium experiments werevery conclusive with respect to the effects of pCO2

on bioerosion rates of single species or microbialcommunities (Tribollet et al. 2006, 2009, Wisshaket al. 2012, Reyes-Nivia et al. 2013), we only havevery little evidence from natural environments atthe community level and over longer periods of time(but see e.g. Fabricius et al. 2011). Therefore, furtherstudies are required to understand future trendsof bioerosion in a larger context and how they willaffect reef carbonate budgets in different environ-ments.

Chemical versus mechanical bioerosion in Clionaorientalis

Sponge bioerosion is a process involving chemicaletching that leads to the mechanical removal ofsponge chips which are then expelled (e.g. Schön-berg 2008). Compared to our value of 9% chemicalbioerosion in Cliona orientalis, a considerable rangeof ratios between contributions of chemical andmechanical sponge bioerosion have been reported inthe literature. Warburton (1958) estimated that <10%of carbonate is dissolved by Cliona cf. celata fromEastern Canada. Based on SEM analyses studyingPione lampa (de Laubenfels, 1950) from Bermuda,Rützler & Rieger (1973) calculated that only 2 to 3%of the carbonate goes into solution, which is the low-est rate reported to date. Zundelevich et al. (2007)simultaneously measured chemical and mechanicalerosion in Red Sea Pione cf. vastifica (Hancock,1849), and found that chemical dissolution was thedominant mode of bioerosion, contributing 75% tothe total rate. More recently, Nava & Carballo (2008)refined the method introduced by Zundelevich et al.

124

Wisshak et al.: Global change impacts on sponge bioerosion

(2007), and quantified chemical bioerosion at 29% inthe Mexican Pacific Cliona vermifera Hancock, 1867,and 12% in Cliona flavifodina Rützler, 1974, respec-tively. We argue that the strong variation in the contribution of chemical dissolution may partly beinfluenced by the pH of the surrounding water, con -sidering that according to Doney (2006), the 2 siteswith the experi mentally determined highest ratios ofchemical bioerosion—the Red Sea and the MexicanPacific (Zundelevich et al. 2007, Nava & Carballo2008)—have comparatively acidic seawater (ca. pH7.9 according to Doney 2006), while the classic stud-ies with low ratios of chemical bioerosion were con-ducted in less acidic areas (Warburton 1958, Rützler& Rieger 1973; around pH 8.1 according to Doney2006). This theory may be supported by observationsby Schönberg (2008), who noticed significant differ-ences in sponge scar morphologies comparing thewidth of the groove etched to remove the spongechips.

As for Cliona orientalis, the increase in the chemi-cal bioerosion rate with increasing pCO2 is muchmore pronounced than the increase of total bioero-sion determined for the same species and under sim-ilar experimental conditions (see Fig. 5 in Wisshaket al. 2012). This is, however, not a contradiction, butmerely the reflection of a shift in the contribution ofchemical versus mechanical mode of sponge bioero-sion. Keeping in mind that our earlier experiment(Wisshak et al. 2012) was conducted in a light/darksetting over a longer period of time and in a flow-through situation, the marked difference in theslopes of the increased bioerosion capacities relativeto a change in pCO2 is not affected by experimentaldifferences, and thus suggests that it is primarily thechemical bioerosion that increases. It is not surpris-ing that it is the chemical etching which is directlyaffected by a more acidic environment, because theprocess is acting at the cell-substrate interface,where ambient conditions have to be overcome.Mechanical bioerosion re mained comparatively con-stant with pCO2 as evidenced by the nearly constantvertical offset between the 2 regressions in Fig. 7, butthe absolute ratio of mechanical versus chemical ero-sion needs to be interpreted with caution when com-paring the 2 separate experiments (see ‘Results: Bio-erosion rates’).

While an increase in etching efficiency with in -creased pCO2 would be expected to result in themechanical liberation of proportionally more spongechips, a different effect comes into play: our data suggest that not only will the number of sponge chipsbe increasing, but a substantial part of the enforced

etching capacity may be directed towards an in -crease in the width of the etching grooves (i.e. theetched crevice between the substrate and the chip),resulting in a reduced size of the chips under moreacidic conditions, if assuming that maximum chipdiameter is restricted by cell properties (Rützler &Rieger 1973). Therefore, while global warming mayhave little effect within the physiological tolerancelimits of the sponges, OA will significantly acceleratetheir bioerosion rates, which we hypothesise willlead to only a slight increase in the production ofsponge-generated sediment and a reduction in grainsize.

Acknowledgements. Fieldwork at Orpheus Island was onlypossible with the supreme logistic support of various AIMSstaff members including M. Donaldson and D. Stockham,and the OIRS team S. Kelly, H. Burgess, P. Lawn, R. Wiley,and P. Costello, and greatly profited from the commitment ofthe scientific volunteers R. and D. Wisdom, G. Ford, and A.Ulrich. Statistical analyses significantly benefitted fromadvice provided by K. Murray, J. Russel, and M. Thums,University of Western Australia, and M. Carreiro-Silva, Uni-versity of the Azores. We are indebted to the thorough in-depth reviews provided by the anonymous referees. Thisstudy was financially supported by the Deutsche Forschungs -gemeinschaft (Grant Fr 1134/19), co-funded by the AIMS.

LITERATURE CITED

Acker KL, Risk MJ (1985) Substrate destruction and sedi-ment production by the boring sponge Cliona caribbaeaon Grand Cayman Island. J Sediment Petrol 55: 705−711

Anderson MJ (2005) PERMANOVA: a FORTRAN computerprogram for permutational multivariate analysis of variance. Department of Statistics, University of Auck-land. www.stat.auckland.ac.nz/~mja/prog/ PERMANOVA_UserNotes.pdf

Anderson MJ, Gorley RN, Clarke KR (2008) PERMANOVA+for PRIMER: guide to software and statistical methods.PRIMER-E, Plymouth

Andersson AJ, Mackenzie FT, Gattuso JP (2011) Effectsof ocean acidification on benthic processes, organisms,and ecosystems. In: Gattuso J-P, Hansson L (eds) Oceanacidification. University Press, Oxford, p 122−153

Atkins P, De Paula J (2009) Physical chemistry, 9th edn. WHFreeman, New York, NY

Bower CE, Bidwell JP (1978) Ionization of ammonia in sea-water: effects of temperature, pH, and salinity. J Fish ResBoard Can 35: 1012−1016

Chisholm JRM, Gattuso JP (1991) Validation of the alkalinityanomaly technique for investigating calcification and pho -to synthesis in coral reef communities. Limnol Oceanogr36: 1232−1239

De’ath G, Lough JM, Fabricius KE (2009) Declining coralcalcification on the Great Barrier Reef. Science 323: 116−119

Dickson AG (1990) Standard potential of the reaction: AgCl(s) + 1/2 H2(g) = Ag(s) + HCl(aq), and the standardacidity constant of the ion HSO4

- in synthetic seawaterfrom 273.15 to 318.15 K. J Chem Thermodyn 22: 113−127

125

Aquat Biol 19: 111–127, 2013

Dickson AG, Millero FJ (1987) A comparison of the equi -librium constants for the dissociation of carbonic acid inseawater media. Deep-Sea Res 34: 1733−1743

Dickson AG, Afghan JD, Anderson GC (2003) Referencematerials for oceanic CO2 analysis: a method for the certification of total alkalinity. Mar Chem 80: 185−197

Doney SC (2006) The dangers of ocean acidification. Sci Am294: 58−65

Duckworth AR, Peterson BJ (2013) Effects of seawater tem-perature and pH on the boring rates of the sponge Clionacelata in scallop shells. Mar Biol 160:27–35

Erez J, Reynaud S, Silverman J, Schneider K, Allemand D(2011) Coral calcification under ocean acidification andglobal change. In: Dubinsky S, Stambler N (eds) Coralreefs: an ecosystem in transition. Springer, New York,NY, p 151−176

Fabricius KE, Dommisse M (2000) Depletion of suspendedparticulate matter over coastal reef communities domi-nated by zooxanthellate soft corals. Mar Ecol Prog Ser196: 157−167

Fabricius KE, Langdon C, Uthicke S, Humphrey C and oth-ers (2011) Losers and winners in coral reefs acclimatizedto elevated carbon dioxide concentrations. Nat ClimChange 1: 165−169

Fang JKH, Schönberg CHL, Kline DI, Hoegh-Guldberg O,Dove S (2013) Methods to quantify components of theexcavating sponge Cliona orientalis Thiele, 1900. MarEcol 34: 193−206

Francis-Floyd R, Watson C, Petty D, Pouder DB (2009)Ammonia in aquatic systems. #FA16, University ofFlorida IFAS Extension, Gainesville, FL

Glynn PW (1997) Bioerosion and coral reef growth: adynamic balance. In: Birkeland C (ed) Life and death ofcoral reefs. Chapman & Hall, New York, NY, p 68−95

Hansen HP, Koroleff F (1999) Determination of nutrients. In: Grasshoff K, Kremling K, Ehrhardt M (eds) Methods ofseawater analysis. Verlag Chemie, Weinheim, p 159−228

Hill M, Hill A (2012) The magnesium inhibition and arrestedphagosome hypotheses: new perspectives on the evolu-tion and ecology of Symbiodinium symbioses. Biol RevCamb Philos Soc 87: 804−821

Hoegh-Guldberg O, Mumby PJ, Hooten AJ, Steneck RS andothers (2007) Coral reefs under rapid climate change andocean acidification. Science 318: 1737−1742

Holmes RM, McClelland JW, Sigman DM, Fry B, PetersonBJ (1998) Measuring 15N−NH4

+ in marine, estuarine andfresh waters: an adaptation of the ammonia diffusionmethod for samples with low ammonium concentrations.Mar Chem 60: 235−243

Hoppe CJM, Langer G, Rokitta SD, Wolf-Gladrow DA, RostB (2012) Implications of observed inconsistencies in car-bonate chemistry measurements for ocean acidificationstudies. Biogeosciences 9: 2401−2405

IPCC (2001) Climate change 2001: the scientific basis. Con-tribution of Working Group I to the third assessment re -port of the Intergovernmental Panel on Climate Change.Cambridge University Press, Cambridge

Jacques TG, Pilson EQ (1980) Experimental ecology ofthe temperate scleractinian coral Astrangia danae. I. Partition of respiration, photosynthesis and calcificationbetween host and symbionts. Mar Biol 60: 167−178

Kroeker KJ, Kordas RL, Crim RN, Singh GG (2010) Meta-analysis reveals negative yet variable effects of oceanacidification on marine organisms. Ecol Lett 13: 1419−1434

Kroeker KJ, Kordas RL, Crim RN, Hendriks IE and others

(2013) Impacts of ocean acidification on marine or -ganisms: quantifying sensitivities and interaction withwarming. Global Change Biol 19: 1884−1896

Lynch TC, Phlips EJ (2000) Filtration of the bloom-formingcyanobacteria Synechococcus by three sponge speciesfrom Florida Bay, USA. Bull Mar Sci 67: 923−936

Maldonado M, Giraud K, Carmona C (2008) Effects of sedi-ment on the survival of asexually produced spongerecruits. Mar Biol 154: 631−641

Maldonado M, Ribes M, van Duyl FC (2012) Nutrient fluxesthrough sponges: biology, budgets, and ecological impli-cations. In: Becerro MA, Uriz MJ, Maldonado M, Turon X(eds) Adv Mar Biol 62. Advances in sponge science: physiology, chemical and microbial diversity, biotechno -logy. Elsevier, London, p 113−182

Mehrbach C, Culberson CH, Hawley JE, Pytcowicz RM(1973) Measurement of the apparent dissociation con-stants of carbonic acid in seawater at atmospheric pres-sure. Limnol Oceanogr 18: 897−907

Meinshausen M, Smith SJ, Calvin K, Daniel JS and others(2011) The RCP greenhouse gas concentrations and theirextensions from 1765 to 2300. Clim Change 109: 213−241

Nava H, Carballo JL (2008) Chemical and mechanical bio-erosion of boring sponges from Mexican Pacific coralreefs. J Exp Biol 211: 2827−2831

Osinga R, Tramper J, Wijffels RH (1999) Cultivation of marine sponges. Mar Biotechnol (NY) 1: 509−532

Pandolfi JM, Connolly SR, Marshall DJ, Cohen AL (2011)Projecting coral reef futures under global warming andocean acidification. Science 333: 418−422

Perry CT, Spencer T, Kench PS (2008) Carbonate budgetsand reef production states: a geomorphic perspectiveon the ecological phase-shift concept. Coral Reefs 27: 853−866

Pierrot D, Lewis E, Wallace DWR (2006) MS Excel programdeveloped for CO2 system calculations. ORNL/CDIAC-105a. Carbon Dioxide Information Analysis Center, OakRidge National Laboratory, US Department of Energy,Oak Ridge, TN

Reyes-Nivia C, Diaz-Pulido G, Kline D, Hoegh-Guldberg O,Dove S (2013) Ocean acidification and warming scenar-ios increase microbioerosion of coral skeletons. GlobalChange Biol 19: 1919−1929

Reynaud S, Leclercq N, Romaine-Lioud S, Ferrier-Pagès C,Jaubert J, Gattuso JP (2003) Interacting effects of CO2

partial pressure and temperature on photosynthesis andcalcification in a scleractinian coral. Global Change Biol9: 1660−1668

Rodolfo-Metalpa R, Houlbrèque F, Tambutté É, Boisson Fand others (2011) Coral and mollusc resistance to oceanacidification adversely affected by warming. Nat ClimChange 1: 308−312

Rosell D (1993) Effects of reproduction on Cliona viridis(Hadromerida) on zooxanthellae. Sci Mar 57: 405−413

Rützler K (2002) Impact of crustose clionid sponges onCaribbean reef corals. Acta Geol Hisp 37: 61−72

Rützler K, Rieger G (1973) Sponge burrowing: fine structureof Cliona lampa penetrating calcareous substrata. MarBiol 21: 144−162

Schönberg CHL (2000) Bioeroding sponges common to thecentral Australian Great Barrier Reef: descriptions ofthree new species, two new records, and additions to twopreviously described species. Senckenb Marit 30: 161−221

Schönberg CHL (2001) Small-scale distribution of GreatBarrier Reef bioeroding sponges in shallow water.

126

Wisshak et al.: Global change impacts on sponge bioerosion

Ophelia 55: 39−54Schönberg CHL (2002a) Sponges of the ‘Cliona viridis com-

plex’ — a key for species identification. Proc 9th Int CoralReef Symp, Bali 1: 295−299

Schönberg CHL (2002b) Substrate effects on the bioerodingdemosponge Cliona orientalis. 1. Bioerosion rates. MarEcol 23: 313−326

Schönberg CHL (2006) Growth and erosion of the zooxan-thelate Australian bioeroding sponge Cliona orientalisare enhanced in light. Proc 10th Int Coral Reef Symp,Okinawa, p 168−174

Schönberg CHL (2008) A history of sponge erosion: frompast myths and hypotheses to recent approaches. In:Wisshak M, Tapanila L (eds) Current developments inbioerosion. Springer, Heidelberg, p 165–202

Schönberg CHL, Ortiz JC (2009) Is sponge bioerosionincreasing? Proc 11th Int Coral Reef Symp Fort Laud-erdale, USA, p 520–523

Schönberg CHL, Shields G (2008) Micro-computed tomog-raphy for studies on Entobia: transparent substrate ver-sus modern technology. In: Wisshak M, Tapanila L (eds)Current developments in bioerosion. Springer, Heidel-berg, p 147−164

Schönberg CHL, Suwa R (2007) Why bioeroding spongesmay be better hosts for symbiotic dinoflagellates thanmany corals. In: Custódio MR, Lôbo-Hajdu G, Hajdu E,Muricy G (eds) Porifera research: biodiversity, innova-tion and sustainability. Museu Nacional Rio de Janeiro,p 569−580

Schönberg CHL, Wilkinson CR (2001) Induced colonizationof corals by a clionid bioeroding sponge. Coral Reefs 20: 69−76

Schönberg CHL, Suwa R, Hidaka M, Loh WKW (2008)Sponge and coral zooxanthellae in heat and light: pre-liminary results of photochemical efficiency monitoredwith pulse amplitude modulated fluorometry. Mar Ecol29: 247−258

Sipkema D, Snijders APL, Schroën CGPH, Osinga R, Wijf-fels RH (2004) The life and death of sponge cells.Biotechnol Bioeng 85: 239−247

Smith SV, Key GS (1975) Carbon dioxide and metabolism inmarine environments. Limnol Oceanogr 20: 493−495

Stoll HM, Ruiz-Encinar J, Garcia-Alonso JI, Rosenthal Y,Klaas C, Probert I (2001) A first look at paleotemperature

prospects from Mg in coccolith carbonate: cleaning tech-niques and culture measurements. Geochem GeophysGeosyst 2: 1047, doi: 10.1029/2000GC000144

Tribollet A, Golubic S (2011) Reef bioerosion: agents andprocesses. In: Dubinsky Z, Stambler N (eds) Coral reefs: an ecosystem in transition. Springer, Berlin, p 335−449

Tribollet A, Atkinson MJ, Langdon C (2006) Effects of ele-vated pCO2 on epilithic and endolithic metabolism ofreef carbonates. Global Change Biol 12: 2200−2208

Tribollet A, Godinot C, Atkinson M, Langdon C (2009)Effects of elevated pCO2 on dissolution of coral carbonatesby microbial euendoliths. Global Biogeochem Cycles 23: 1−7, doi: 10.1029/2008GB003286

Tribollet A, Radtke G, Golubic S (2011) Bioerosion. In: Reitner J, Thiel V (eds) Encyclopedia of geobiology.Springer, Berlin, p 117−134

van Soest RWM, Boury-Esnault N, Hooper JNA, Rützler Kand others (2013) World Porifera Database. Available at: www.marinespecies.org/porifera

Veron JEN (2011) Ocean acidification and coral reefs: anemerging big picture. Diversity 3: 262−274

Warburton FE (1958) The manner in which the spongeCliona bores in calcareous objects. Can J Zool 36: 555−562

Ward-Paige CA, Risk MJ, Sherwood OA, Jaap WC (2005)Clionid sponge surveys on the Florida Reef Tract sug-gest land-based nutrient inputs. Mar Pollut Bull 51: 570−579

Weisz JB, Massaro AJ, Ramsby BD, Hill MS (2010) Zooxan-thellar symbionts shape host sponge trophic statusthrough translocation of carbon. Biol Bull (Woods Hole)219: 189−197

Wisshak M, Tapanila L (2008) Current developments in bio-erosion. Springer, Heidelberg

Wisshak M, Schönberg CHL, Form A, Freiwald A (2012)Ocean acidification accelerates reef bioerosion. PLoSONE 7: e45124

Wright P (1995) Review nitrogen excretion: three end prod-ucts, many physiological roles. J Exp Biol 198: 273−281

Zeebe RE, Wolf-Gladrow D (2001) CO2 in seawater: equilib-rium, kinetics, isotopes. Elsevier, Amsterdam

Zundelevich A, Lazar B, Ilan M (2007) Chemical versusmechanical bioerosion of coral reefs by boring sponges—lessons from Pione cf. vastifica. J Exp Biol 210: 91−96

127

Editorial responsibility: Paul Sammarco, Chauvin, Louisiana, USA

Submitted: March 26, 2013; Accepted: July 19, 2013Proofs received from author(s): September 12, 2013